As computing power increases numerical weather prediction models are being run at increasingly higher resolutions. The validity of higher resolution NWP depends on the validity of parameterizations used to represent sub-grid scale processes. As the grid size changes, so does the classification of what processes are sub-grid scale, and thus the applicability of the parameterizations. The varying size of eddies within the convective boundary layer makes their parameterization difficult. The effect of the smaller eddies can be represented by simple K-Theory which employs the use of the local gradient of the transported variable. However this type of turbulence closure allows the development of unrealistically deep superadiabatic layers in robust convective situations. Large convective eddies can transport quantities through the depth of the boundary layer, often counter to the local gradient, and thus a nonlocal turbulence closure is needed. A nonlocal turbulence parameterization was added to the University of Wisconsin Nonhydrostatic Modeling System (UW-NMS) to improve simulations of the convective boundary layer observed during Lake-ICE. Simulations were performed with and without the nonlocal closure, at both cloud resolving and coarser resolutions. At cloud resolving scales, the largest convective eddies approached resolution, and the nonlocal turbulence parameterization was shown to have a gray area. This gray area is similar to that of cumulus parameterization, where the convective eddies can not be explicitly resolved, yet are too large to be parameterized effectively.